 Thanks, everyone, for attending this session. As Trishan mentioned, my name is Balaji Ethirajaloo. And today, I'm going to talk about, I know since morning you probably talked about Fido, little bit deep dives. There are some more deep dives coming further. But I'm going to connect some dots here at the highest level from the 5G. Then we'll dive into a little bit of orchestration, network automation, network slicing. And for different types of 5G use cases, why do we need low latency, high throughput, universal data plane? So you will see why it is that it's very important. So as you see, this is a high level. I'm going to go through this. 5G overview, some of the use cases, orchestration, network slicing, and automation, and etc. Folks, we are on a brink of an extraordinary revolution here in terms of pickup of users and IoT devices. As you see in the picture at the bottom, it took almost 100 years to connect about a billion people, basically the fixed telephony, right? Fixed technology. Then it only took about 25 years to connect 5 billion users, predominantly mobile technology. Now, also it will take, by 2020, we estimate to have about 26 billion devices, predominantly Internet of Things, IoT or M2M devices. All will be connected in some in 4G, some in 5G networks. But we are on a well on our way to our original prediction. Ericsson actually projected those figures, 50 billion devices. So maybe it's beyond, or at least our projection is up to 50 billion devices. So that is the, you know, how the device ecosystem will move on. Next, before I talk about 5G, I just want to give, go back a little bit about, I'm not sure everybody's aware of how the mobile technologies evolved. So we started with 1G, I started with 1G, and then 2G, 3G, 4G and 5G. So first generation mobile technology was predominantly purely voice. That's all we did. So in fact, 1992 is the year I got my first cell phone. I got my car phone that was, the battery was like our briefcase, the whole car phone. And then my cell phone was that big, with a huge orange light and big buttons. And I was so excited, the use case was just voice, right? First time I was liberated that I can speak to my parents from my car. That was a very joyful occasion. Made an international call from my cell phone, 92, but today it's laughable, right? The kind of use cases we have. So the 2G, so that was all voice on the one first generation. And the second generation was, we introduced SMS. Famously people call texting now, right? So it's SMS short messaging service. That was a big deal. So we could do, we do some texting, we did some texting. And the third generation was voice, texting, plus the data. We started giving a little bit of data speed on the third generation. Even 128 kilobytes was a big deal. But we started. So it was 2.5G, 7G and 3G. The 4G came along with, you know, after 3G, LTE, long-term evolution. With 4G, it's all exactly like 3G, but much, much, much higher speed. That's 4G. 5G is going to be all completely, a big game changer. Completely different. It's not just about, you'll see. I'm going to cover a lot about 5G. So one of the key things in 5G, oh, one more important thing. In 5G, up to 4G we talked about, you know, mobile core and, you know, the access, radio access network. In 5G, it's much, much more than that. So you're talking about radio access. You're talking about, you know, SDN control, advanced automated transport networks and the mobile core, which is virtualized, right? On top of that, it's not just the networks alone, but the thing that will make the network programmable and automated way, right? So we're going to use management and orchestration and very advanced analytics. So all will, as part of the technology evolution, we're going to use the best technologies. By the way, if you go back to the first generation to now the fifth generation of mobile technologies, telecommunication always we try to use the best technologies that are available at that point in time. Always, pretty much, we use. Now, I see actually, as part of the 5G, it's just the beginning, right? We may have more technologies in the next two, three years. For example, I will say for data integrity, security, blockchain will be used, for example. So everything evolution, the technologies will start incorporating in 5G. What will happen is we're going to create such a network, which is high agility, automated, that could actually create new use cases. I may cover some use cases today, but there are use cases that we're not even aware that will pop up. Just remember this. I always think about this. Without the 4G high speed connectivity, imagine if I land in LAX or SF4, waiting for Uber. Now we are looking at 4G, man, the car driver is coming so close, right? We're all watching that. In 2G, that was probably not almost impossible. It's very slow, probably data is, you cannot even, it'll take a long time. So just an example, right? So we may see much, much different, more use cases coming out of 5G. So end result, so operators, the way they do business today, they will enter into new segment, various different new models, business models that will create different sorts of revenue models as well. Okay, so that's about 5G in essence. I want to touch upon what are the key challenges in terms of with respect to 5G. So operators and different types of operators, they will expect very, very highly complex services in terms of extraordinary amount of devices we just saw in the first slide, 50 billion devices, maybe even say 20 billion devices. So highly scalable network to take care of many, many billions of devices. Scalability is key. Then also some of the services require extremely low latency, right? So networks need to build for that because all these things will run on our network. And then bandwidth requirements and moving things different areas, you know, cloud distributed cloud, centralized cloud. So complex services will be expected that is given, those are the challenges. Going back to the next one is the key area of challenges is agility. So just to give people have different things, you know, a thought process about agility. Just give from the operator perspective, from the business process perspective. Agility is basically the ability as soon as my customer, my end user calls me enterprise or whoever, the ability for me to actually process that requirements and request to create a new product. That's the time to market. Now it may take three months or two months depending on the service, but I may expect in 15 minutes as soon as I make a call, I need the service ready and deployed. So creation of the service, design and development of the service, time to market. Then the fulfillment has to be really, really fast, right? So that is agility. So it is expected, a lot of agility is expected in the 5G use cases. Of course, complex services, best of the technology I need, great agility. I wanted a very low cheap price, right? So cost is a key factor as well. Yeah, because we cannot just scale up the way we scale up today for a 50 billion device. Those devices will not pay a lot of money and the output will be much lower. Just an example. Now to deliver, to meet all those challenges, what are the key ingredients, right? As part of the 5G, I covered a little bit. Orchestration will play a great game. You know, it will be a good game changer on that. So orchestration and automation is key. Programmability of the network, being network being very flexible is very key. Modularity of the network. Orchestration layer should be modular. Functions should be very modular. And obviously I'll talk about network slice. That's a concept that will come in 5G, meaning in a single network, same physical infrastructure, delivering a logical network slices for different industry in a same network. Okay. I just threw in some use cases. Well, you can think many use cases in the future. It will, you know, will develop much, much more. So the first one, as you see, the full-length HD movie. Probably expected to be downloaded in a couple of minutes. Full length. I'm not sure how long it will take now. Then another interesting thing is the car use case, right? Automative cars. For example, using 4G, we've done some studies using 4G. The decision is made based on all the technical information that you receive from the car. Goes to the cloud. Decision is made. Signal is sent back to the car. Okay, it's time for you to apply the break. That process in 4G will take, because of the latency, something distance-wise about 4.5 feet. Maybe it's okay. But in 5G, that's reduced to 200 inches. So decision can be made much, much later. See, the latency is very, very low. It is expected to be very low latency. A network has to be built for that. Obviously, I don't know, 4GB capacity, much higher, even 20GB, whatever the high capacity 5G is going to give us. So we will replace. It will be equal to fiber or above fiber. So we can technically have, in a fixed access, we can have the 5G pretty much. Another critical factor for Internet of Industrial Internet, or IoT, is key this one is, because of the transmission requirements. Power requirements will be very, very low, right? So what will happen if I have a remote device somewhere sitting there in mountain or somewhere. So I cannot be going there changing battery all the time, every six months. 5G will make sure that because of the power consumption needs, the battery life will be extended 10x than what we have today. So that will actually create more devices in the ecosystem. So that is very critical. So obviously the virtual reality AR will have a tremendous boost, those use cases in 5G. So these are the typical, you know, I just showed you a few use cases. But again, now going back to, so I talked about network slicing. So these are the beginning slice on the network slice. So you have the single physical infrastructure, single network. So that network will logically create network slices on the physical infrastructure. So different services will be built logically separated for the service. Each network slice may belong to one service instance. So it will be delivered from different cloud towards different vertical industries as you see in that. Okay, let's go to the next one. Now going back, well, the journey has already started. You know, we started virtualizing the network function. Somewhere in 2014, we started talking all of, you know, we had a lot of parks and trials. Now it's already in production environment. Many of the network functions already in production. So virtualization is first step. You need to have the virtualized network functions, virtualized network networks, right, step one. Then the networks, depending on some of them are legacy networks. So you need software defined control of the network. So SDN plays a key role along with virtualization. The next is key part is distributed cloud. Once it is virtualized, here is a situation, right? Some of my functions, I'll keep them in a centralized data center, for example. Then depending on the user needs, right, may require a very low latency. For example, I may move my H.264 codes, you know, media resource, this codecs. I may move much, much closer to the user, right? So those things as a VNF or in, you know, either VM or in container will stay much, much closer to the distributed local data center. Just a typical example. So some functions may move to central, some functions may move towards the end user. So obviously the network slicing, I have a couple of slides I'll get into detail about network slicing. Anyway, to connect all of them, to hold them together as a glue on the top, you need a management and orchestration and analytics layer. So I want to touch upon little bit on the orchestration. So a lot of people think about orchestration in, they all have a different opinion, different ideas about orchestration. So from a pure technologies perspective, I just want to show, just peel the orchestration a little bit, so it's much easier to follow. I just divided, again, it's nothing about the products or open source. I'm talking purely technologies perspective. Oh, the orchestration will work. The lowest level, much closer to infrastructure. Excuse me. So this layer basically manages, orchestrates your compute storage network resources. Typically we're in open stack. I'll say heat orchestrator belongs there, just to give an example. Then about that, you have an orchestration layer that is for the whole networks. So I can orchestrate my L3 VPN, my underlay, overlay, and I'll also orchestrate my network functions. Very important, this is not only the virtualized network functions. PNF is very important, physical network functions. Some of the functions may ever, they may not even move to the virtualized world. They may have to stay as PNF. So I may have to orchestrate all together. So that's that layer. It's mostly operational layer or a technology layer, virtualization network layer. On top of that, what I call end-to-end service orchestration, which is basically a business layer, which is pretty much tied with your product catalog, service catalog in the business layer. So that's where I define my end-to-end service. For example, I may define end-to-end voltage use case, voice over LTE use case, for example, or virtual CPE use cases. So that is all defined on the service layer. So from the service layer, it will decompose into different, you know, workflows. And these workflows will go to, you can go to different, either to NFEO or to an NMS or to an SDN controller. So that's the key part. Then if you talk about the middle part, which will include, obviously, which is very, very key, because this is the networks that will include, you may have the, you know, transport for transport, you will have the multi-layer van controller and also the L3, L2, VPN definitions. All of that will be handled by the central layer of the VXLAN connectivity, intra-data-centered connectivity, and also inter-data-centered connectivity. And the VNF itself will all be orchestrated in the middle layer. Also, I will add something on the middle layer, which is also critical, which is, for me, the second and third layer will also apply policy and analytical decisions. And so we'll apply closed-loop automation. We don't need to wait for the closed-loop automation to go all the way to the service layer. You should be able to make sure actually in the VIM layer itself you should be able to close something, make sure infrastructure is very mobile and, you know, flexible all the time. So if you take the middle layer, middle layer gets all the performance data, the counters, alarms, and events from the different elements, and also it will correlate against the network services and using analytics principle, it can take action and close the loop and solve the issues at that layer. Obviously, the service orchestration layer will actually monitor the mean opinion score end-to-end how I'm delivering my services for a specific SLA that I already defined. So that's the top layer. So you saw that, how the flexibility you're going to achieve because of that. Okay, here I'm talking about, you know, same thing about network slices, what I explained before. So you have the physical resources based on that the network function will have on the left side. It will have physical resources plus also some logical resources. Then as part of the network slice, if you see OSS-BSS itself, part of the OSS-BSS itself is part of the network slice, right? So that's key to note. So obviously, my access transport cloud will be all part of this network slice. They are the resources for the network slice. So if you see on the top layer, there is a business layer, which I talked about service orchestration, that is business layer, and the network governance layer, that's the middle orchestration layer I talked about in the previous slide. So together combined, they will create a specific network slice for a specific use case. In this case, you could see I'll have a separate slice specifically for a mobile broadband, maybe to an enterprise, or an MUNO scenario. Then I may have for a specific industry a slice for that, okay? And also some for massive IoT, I may have a slice for the IoT devices. I just want to give you a little bit more deep dive on network slice, and I'm going to give you a little bit more about the orchestration, NGMN has defined the network slice. This is very critical to understand. The beauty here is if you see the slide I explained about the orchestration different layers, this is pretty much false into that. It is a deal, right? Remember at the beginning of the conversation, I mentioned about flexibility and automation. This is a fundamental. You cannot have the rigid network. So it has to be very flexible. So if you see the network slice itself is very flexible. There is a key, which is basically a network slice blueprint. So that particular network slice will consist of, I showed in the previous slide, the network resources, even the OSS-BSS part of it, access, transport, all of them are defined in that network instance. Now if you see below, a sub-network instance, for example in a mobile core a sub-network instance, possibly a virtual MME, or EPG, could be a sub-network instance. So if you see that, if you see the third one, network slice instance three, that will consist of different sub-network, right? So that's the flavor. So you could have a different sub-network instance connected with the network slice. Then on the top, basically overall each service instance will have one network slice defined for it. But due to some business needs, maybe a small number of subscribers, whatever the reason, policy decision by the operator, that business decision, they could create one network slice instance. They could run two services on that, today. But later, they can, you know, define themselves, oh, you know what, under certain conditions, certain policy, I may have to change, so I'll spin up a new network slice and move the service instance five to the network slice five. So that possibility is very flexible. Obviously at the bottom layer is all the resources, physical infrastructure, all of them. So you could see how the flexibility is there, right? Just to define the network slice in 5G. So this is a little bit, you know, same thing, but little bit detailed. So if you see the business orchestration or the service orchestration layer is for the business flexibility. The network flexibility is created by the network layer orchestration. As you see here, the characteristics are different between the two layers. The network governance layer or the network orchestration provides operational capabilities. Technically, the technical characteristics that you can adjust for the network slice, for example, different features in a network slice and different characteristics, maybe a very low latency or a high latency, all these types of characteristics. Then distribution, you may have to, you want to have some of the network slice functionality in closer to the user, some closer to the central data center. So that's distribution. So all these characteristics built in a blueprint for that layer orchestration, the network layer orchestration. Isolation versus sharing, we talked about, right? Two services in a single network slice. It's sharing. But you want to, for a certain reason, maybe a specific customer, they don't want to be, the SLA clearly demands, I want my own slice, nobody's touching it, I want secured and delivered to me. Could be isolation. Connectivity is key again, obviously. Scale is very important. Some of the network slice I want to create for a highly scalable infrastructure, highly scalable use cases. So this is the network level characteristics, right? Then combine that with the business layers orchestration, which is service layer. There you can see different parameters, provides a flexibility there. So for some use cases, for some network slice, different business model altogether, right? I cannot get into all the details, because of 40 minutes, but different business model, then also different building and charging models could be. And then obviously the SLA, some really requires the best SLA. So you need to create, based on that, you will create a network slice. Obviously the end result is you could see different slices. I just want to touch upon automation. Remember I mentioned at the beginning where the analytics and automation play a role in this. Take an example here. There is a network slice for a mobile broadband. Slice is created. It's working. And due to specific policies, both the network policy, as well as the network as a service governance you see on the top. That's basically the service layer orchestration. So we'll be constantly monitoring. That layer will be monitoring the mean opinion score, maybe the SLA, what not. So when it realizes, okay, it could be dynamic or it could be pre-programmed. So take it dynamic. If they see, okay, some violation of the SLA, whatever the reason may be from the network, they could take an action and send in information to the network layer, network slice governance layer to spin up a new network slice and then move maybe all users or some users to the new network slice. So this is one use case. There could be another use case where, you know, Levi Stadium and there is a match going on. I want to instantiate a network slice maybe for two, three hours. We can instantiate, keep the traffic in the slice. Game is over. We close down. Resources are freed up. So all these things are possibilities in 5G, a clear possibility is how we tailor them together to stitch together to make a use case and services. Yeah. Okay. Well, I moved from 5G network slicing automation and then, okay, let's touch upon a little bit about open source and the data plan FIDO, right? I'm sure you've probably seen many FIDO slides since morning. Okay. This is a high-level ecosystem or a landscape of extraordinarily the open source. Okay. Let me step back. At the end there, it's standard. So different standards in our industry that's working together. And then in the left side, you see the different functional layers, right? Operating system where Linux stays, network control, cloud and management, orchestration policy and analytics. And you could see a parallel matching. For example, you see Panda on the top on the analytics. ONAP goes two, three layers. And then you can see open daylight and ONOS on the control layer. Also, you see ONAP and open source mono as well in the orchestration layer. RES, TASCOR, TASCOR basically. So you see all of, you know, different ecosystem, different open source projects, standards in the same ecosystem. I'm going to go to the next slide, which is a little bit coming down more towards network layers, more closer to us. So if you see the middle layer, obviously the FIDO. So you have the open daylight. And then you have, at the bottom, you have obviously next operating systems. On the top you have orchestration layer, ONAP. I'm sure many people touched up on FIDO, but maybe some new people are here. I just want to talk about a little bit one slide about FIDO. Obviously, we talked about 5G and all the use cases, but I mentioned two important words. I need a very high throughput and a low latency system, data plane. That's most critical. Obviously, there are a lot of features we can talk about it. And then a key part on FIDO is it can work on bare metal server. Obviously as a virtual machine you can deploy or in containers you can deploy. It's also available as a VNF. You can use a routing L2, L3 function in a VNF for routing needs. Not in the NFEA, not in the data plane, but as a routing needs as a VNF. Obviously it can work in different chip vendors here, Intel, ARM, and one more. And future whatever comes out we are very open. If you see the FIDO basic idea on the last end of the slide there, Network IO is DPDK where the data packets are coming in and then it's processed in VPP, vector processing at the middle. And the top layer is the management layer where we expose using any comb and interwork with different layers. I just want to show this one slide where we are working on some of the integrations with FDIO. One is integration with Open Daylight. So there is a proof of concept or a trial our guys are doing, what we call network project, network virtualization project. So that is to integrate Open Daylight with Honeycomb, using Honeycomb agent on FIDO. Then also for flexibility you could see we are also integrated with OpenStack. It's integrated with FIDO. Okay. Couple of more slides. So here I talked about FIDO a little bit, the data plane, what is needed in this universal data plane to meet the needs of the 5G, the future that's coming. This is actually, there are many predictions out there about the economic potential that we get out of this new industry that we are entering into 5G. This is across the whole industry sectors. Large ecosystem. You could see all the different industries. This is one analyst prediction that is the IoT industry will generate about 1.9 trillion dollar worth of opportunities. This is the entire ecosystem. Telecommunication will be part of it. Networking will be even sub part of that. Just shows the magnitude of the industry that we will deal with that. Last close out before the Q&A session. Orchestration in 5G, the orchestration, automation, network slicing, all will play a major role. Obviously touched upon universal high throughput, low latency, powerful data engine we need that. End of the day packets coming in, packets going out. I can have the beautiful things on the top exposing many things, but my V12 engine doesn't work great is a problem. So that's what I look at FIDO to deliver that. And obviously for operators it is a lot of opportunities on the top line growth. So we'll have different business segments opened up for them. By the way, another interesting thing, lack of time. There are so many business cases already happening in 4G, but it will be even more enormously big in 5G. We are delivering services to the farming community, to the wine industry, so many things. So there's tremendous saving potential across the board, even the energy sector. So that's a great opportunity for the operator. And many of the services for example, today operators deliver services to the end user, right? In 5G, starting even in 4G, there will be multiple industry coming together to just deliver one service. For example, the wine industry I was talking about, so it's a winemaker, farmer, he's a final guy who receives the service, but it's a telecom company involved, the vendors are involved, multiple other middle companies are involved, just to deliver one service. So services will flow through many channels. A business model will be totally different. So, yeah. And obviously, network agility is very key. Yeah, I'll stop it here and please question, answer, feel free to ask any questions. Oh, yeah. Yeah, come on, brave enough, ask one question. It's a quiet crowd. Very quiet. Let me ask you a question, guys. I still have some time. Sorry, somebody's asking. Question in the back. You have the mic? Yeah, I've got the mic. So you talked a lot about network agility and orchestration and all that kind of goodness. In what sense is that actually tied to 5G as opposed to being equally deliverable on 4G? Okay, great question. So I'll summarize the question just to follow up. So you're asking me how the orchestration and also automation you said, network slicing, how it is related to 5G, but it's already there in 4G, some level, right? Yeah, I was just trying to understand in your view how much of this kind of agility could not be delivered on 4G but requires 5G? Yeah, absolutely or absolutely right. So as I mentioned in the beginning of the slide, so many of the technologies are evolving. So by the time 4G comes, you will have obviously the radio capacity, a lot of standards are working to enhance the radio capacity with different technology. That's going to happen, right? Step one. But in 4G itself, that's very important. We are transforming already network function virtualization is already started, which is fundamentally one of the key corner stones of that 5G. But it's happening in 4G already, right? Obviously some network functions are virtualized 10%, 20% but given in say expecting 2018, 19 when the radio technology comes in. By the time I expect many big operators will have majority of their network functions are virtualized. So two more years, right? So virtualization will be in full swing. Transport capability will be very programmable, multi-layer van controller, a dynamicity on the transport layer. Obviously some legacy will be there. Then also the advances made in analytics. Also I talked about blockchain. So for some of the use cases, we need to have some data integrity. Very key especially for the IoT security because the legacy devices doesn't have any security. If I connect them, well, better provide a lot of good security for the devices. Especially the data that I'm transferring should be, you know, the integrity should be good. Yeah. Thank you. I guess I'm next. Okay. So I'm trying to understand how the abstraction of network slicing translates down to what actually happens down the level where I work, which is down near the data plane. So my concept of network slicing, are there like horizontal slices sort of like you might have one slice that the big overall network might be AT&T or Verizon or one of those guys delivering lots and lots of bandwidth and then underneath it you might have a content provider that has a slice that uses maybe only 30% of that bandwidth or maybe a sub-seller and then underneath that you have multiple slices that offer different services and so on and then underneath that the lowly consumer. Is that kind of the concept that you're talking about here? How does that tie into orchestration? No, you pretty much talk, you know, very good points you made on that. So just to clarify, the question was about, you know, network slicing in different layers. You're right. So as soon as you have the networks virtualized and you're providing also the orchestration programmability and automation capabilities, so you're ready to do the network slice. Obviously, that could be a large network slice for an MUNO scenario. As an operator, I want to actually today we can do the MUNO. Obviously, we've been doing that, right? It's a very rudimentary way. I don't want to get in details what I do in my switches, but a lot of things we do. Very rudimentary way, I can tell you that. But once it is virtualized in a logical way, it makes it much, much easier, much, much faster we can create a network slice. Now going back, a large network slice, then also you could create a network slice for massively scalable, massive IoT use cases, right? I may have 500 million devices in my slice. So that is a requirement for that. Maybe latency may not, doesn't matter, but the scalability is important. So different characteristics. You saw the flexibility for the network governance layer and the business governance layer, right? That flexibility is very key. So I can adjust my, you know, both business level as well as the technical characteristics level. Also, I can create a very small network slice for, you know, a few number of users. And I can also, time-wise, I can control, policy-wise, I can control, create a network slice. I was talking about the stadium event, maybe for three hours, and I'll close down. So another interesting thing, actually, I could see, remember at the very beginning, I talked about services and the agility. So remember, today, if my CMO in the operator world, I want to create this nice service, right? We figured out a good service that we can make money on. So I come back to my engineering team, I, you know, discuss with them, CTO group. By the time you need to have, you need to show good business case, I need to make investment, it takes about three months. I don't know how long, even further, to deliver that service. Imagine the service doesn't perform, we will not figure out until we deliver the service, because users have to enjoy it, and see how it grows. In network slicing, with the agility, I don't have to wait all of this. I will introduce a service, users are fine, are not happy. If it doesn't pan out, I'm going to shut the door, remove it. This is, you know, another great advantage of network slice. Thank you. I still have about seven minutes. So, anything open? I thought what you were saying at the beginning about the difference between the 4G networked smart car and the 5G smart car was very interesting. Would you mind talking about that a little bit more? Sorry, I didn't capture this last step. I just asked if you would talk a little bit more about how smart cars will change between 4G and 5G. Absolutely, yeah. So to touch upon again, obviously I just took one small example of application of the break, right? But there are other factors involved on that, which is for me, one is technology to really, for that industry, that particular aspect to take off, I even feel larger thinking from our own humankind society should make certain decisions. Because today if I make an error, it's my human error. I made some error, right? I may make an error not to hit that person, but I to make that person to survive, I may change the direction of the car and may hit this person without getting into details, but as a good faith, I did that because I'm going to hit one of them, right? Possible. In 5G, if you make the machine to do that possible, you can, then there will be questions and the loss has to be created to address that, but you cannot just, okay, that kid, I'm going to hit this older person or whatever, right? Different things. So even when you give to the machine, then there will be, so I see a lot of implications in that area. Not the technology alone, but really in that specific area, I see more of a society making decisions together. Absolutely. But technology is there. Technology will be there. Absolutely. You know, I see one more interesting thing. I'm more safer to be honest with you. I'm a 680 and a 580 in my area and the Bay Area, so many people just, they're all, I'm just watching them, they're all on their cell phone driving all the way. Some watching YouTube videos scary. I don't know how we are running without accident, but imagine self-driving car. We can take the pain away. You'll be on your texting all your life. Let the car take you somewhere. So there will be a situation where probably it's much easier to make, as a society, make those decisions than allowing, because law will not control it, human being, they're going to do that. Whatever law you apply, you will still text or watch video or read the e-mail as well driving. So yeah, there's a big use case eventually to solve the problem. Okay. We have time for one more question. Balaji, do you have any final comments? Yeah. My final comments is I expect we will deliver the network eventually as evolving, different proprietary open source, but it is the use cases I'm very excited to see how as a society we'll create so many use cases that will change our lives forever actually in 5G, starting in 5G. Thank you guys. Thank you.